Friday, July 6, 2012

MLS Goalkeepers Dig the Long Ball (Maybe They Shouldn't)

As most serious FIFA gamers know, you generally want to build from the back - quick and short passes cut up the defense whereas aimless long balls often end up with the opposing team in possession.  The situation is not too different in real life.

I decided to look specifically at Goalkeeper distribution from the mid-week MLS games (7/3-7/4).  I wanted to look at the differences in outcomes between "long" passes (what OPTA calls "launches" and other 40+ yard efforts) and "short" passes.  This seems simple enough.  You would expect a much higher completion rate for the short passes and much lower for those booming goal kicks.  However,  I wanted to take my analysis a step further by looking at the outcomes of each of these possessions.  I used three possible outcomes (many more could have been used): turnover on own half, turnover on other half, or attempt on goal.  Admittedly, this is a small sample size but I think there are still some interesting results, though they are pretty intuitive.

As seen above, those long hopeful boots down the field are completed at a paltry rate of 38%.  It should be further noted that many of these "completions" are to lone attackers who are quickly disposessed even if they are lucky enough to control it.  Of course the converse is that while these short passes are essentially sure things, the defender or midfielder must still find a way to break down the first line of defense, leaving them vulnerable to the dreaded turnover in their own half.  Below find a summary of these findings:

It is not necessarily apparent that one approach is better than another.  Again, this is a small sample size and attempts on goal are not always how you define a successful possession.  One thing does stick out to me:  while the number of turnovers in a team's own half is much higher when using short passes, teams still seem able to advance the ball fairly easily past midfield (81% of the time).  Also,  I should point out that teams that use short GK passes seem to be more comfortable playing this way (and vice versa for non- short passing teams) as can be seen by this chart.  This may also be a reflection of how much defensive pressure the opposing team was exerting:

Also of interest, the differences in approach between goalkeepers can sometimes be quite stark.  The comparison between Troy Perkins vs. San Jose (most long-ball heavy) and Matt Pickens vs. Vancouver (most short-pass heavy) illustrates this.

Troy Perkins

Matt Pickens

What does all this tell us?  Well, not a lot that we didn't already suspect.  If a team has the ability to play from the back, it certainly seems preferable.  This is especially true when holding the lead and looking to kill time.  Goalkeeper distribution is not unlike a punter or a special teams unit in American Football (hear me out, I know it is a ridiculous comparison) in that you're not often going to score off of these plays, but they are important for field position.  And field position is important. 


  1. One thing worth noting is the approach of the other team on GK distribution. In the example of Pickens vs. Vancouver, the Whitecaps sat in, on the road and with excessive heat. This put the onus on the Rapids to break them down. Why boot it long to a bunkered in team when a short pass to center back get's the attack started?

    Would be interesting to see more analysis based on home-away and considering the approach of the other team. Most GKs play long due to press of the opposition, tactical instruction from their coach, or they don't believe there defenders and midfielders have the ability to play out.


  2. I agree that a GK's distribution is really more a reflection of their team, score, home/road, and other factors outside of their control. I also would like to see more analysis/data on this, and in fact I felt a little bad putting this together considering just how small the sample size really is. But, finding the possession outcome for every pass was quite grueling and came out to about 30 minutes of quite tedious work per GK.